Skip to content

Cobalt prices

WordPress added various functions over the summer, and I'm testing the ability to embed graphs here. Why cobalt prices? Cobalt is a critical part of electric vehicle batteries, as cathodes rely upon nickel-chrome oxide crystals of various compositions (eg, along with manganese). It is a potential game-stopper, because enough is used that chrome content affects the overall price of a car. There are other chemistries ... in the lab. However, the path from lab to cost-effective volume production is at a minimum 20 years, and then adding capacity to meet the potential size of the market can (in the automotive case) take an additional two decades.

Now this isn't macroeconomics, but it is certainly a good exemplar of the challenges of turning exciting technologies into commercial realities. That IS of macroeconomic significance. Indeed, in the US context productivity is potentially the single most important contributor to economic growth over the 40-plus years your working life.

This however is a statement that needs both a theoretical foundation – what underlies growth? – and am empirical foundation – what are the numbers in the US context, and can we usefully project them for the next 4 decades? Making predictions is challenging, the moreso if they are about the distant future. Be skeptical about claims that pin that down to a precise number; be thoughtful about whether we can put likely bounds on what will happen, e.g., can we achieve 3% growth?

Source: tradingeconomics.com

If I understand correctly, both these graphs should update automatically – come back in a month to see whether they still reflects Aug 26th data, which for cobalt happened to be the then historic peak of $60,750 per (metric) ton.

6 thoughts on “Cobalt prices

  1. the prof

    I updated the initial post with a graph of productivity. I provided no explanation though of what the units are or otherwise what it represents.

  2. johnsonj20

    Interestingly, Tesla's value in terms of market cap has more than doubled since 2015 even though it only produced 76,230 vehicles in 2016. Tesla's gargantuan stock growth in the past two years is the only valid source of its value surge, because its other statistics in categories like revenue, vehicle deliveries, employees, and profit are relatively dismal compared to those of other major car manufacturers. Despite Tesla's envious stash of investors, its production needs to skyrocket for growth to occur.

    1. the prof

      I've blogged extensively on this (and also have many comments at Seeking Alpha). Tesla is a bubble, pure and simple, a stock price unsupported by fundamentals. It's bonds have "junk" status; it has only made profits where it's sold a year's worth ZEV (pollution) credits in a single quarter. The Model 3 is only just entering production, and cannot make a profit at $35K while Tesla has now lowered Model S and Model X prices two times, and indirectly offers discounts by selling "loaners" (with near-zero mileage). I see no evidence they've lowered their costs, despite hoopla the Nevada battery "Gigafactory" has no impact on materials costs, which dominate battery costs, and are not falling.

      However, I'm not about to "short" Tesla stock. That costs money, and there's no obvious trigger to cause the share price to fall in the next 6 months that isn't already out there (eg, bad quarterly financial reports – Tesla has given guidance that the next one will be worse).

      Anyway, I'm penciled in for a Tesla conference call next week with an [anonymous] investment bank audience.

  3. yuy20

    I'd be interested in seeing what kinds of models are in use now for number projections in the US context and especially in how they deal with possible unexpected events

    1. the prof

      Not sure which projections you have in mind, but you can do a model-based projection, or a (fancy) statistical trend projection. For "known" variance, you can factor in deviations and build a "fan" graph bracketed by high/low values. That in other words uses a model of "unexpected" events. However, the bigger the surprise, the worse a model is likely to perform – by construction statistical models work best for handling deviations near the mean, not outliers. Models may still offer a conceptual framework, but not be very useful empirically.

      Projecting gasoline prices can use data on consumer incomes in the US and elsewhere, for example, but may not be helpful for Harvey – it's a supply interruption in an important region, not a "global" event. And it's likely to be transitory – refineries may already be reopening. If physical shortages are mainly in Texas, will that affect prices in Washington State?

Comments are closed.